Fast, accurate, and racially unbiased pan-cancer tumor-only variant calling with tabular machine learning
Abstract Accurately identifying somatic mutations is essential for precision oncology and crucial for calculating tumor-mutational burden (TMB), an important predictor of response to immunotherapy. For tumor-only variant calling (i.e., when the cancer biopsy but not the patient’s normal tissue sampl...
Main Authors: | , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Nature Portfolio
2023-01-01
|
Series: | npj Precision Oncology |
Online Access: | https://doi.org/10.1038/s41698-022-00340-1 |
_version_ | 1797641509188665344 |
---|---|
author | R. Tyler McLaughlin Maansi Asthana Marc Di Meo Michele Ceccarelli Howard J. Jacob David L. Masica |
author_facet | R. Tyler McLaughlin Maansi Asthana Marc Di Meo Michele Ceccarelli Howard J. Jacob David L. Masica |
author_sort | R. Tyler McLaughlin |
collection | DOAJ |
description | Abstract Accurately identifying somatic mutations is essential for precision oncology and crucial for calculating tumor-mutational burden (TMB), an important predictor of response to immunotherapy. For tumor-only variant calling (i.e., when the cancer biopsy but not the patient’s normal tissue sample is sequenced), accurately distinguishing somatic mutations from germline variants is a challenging problem that, when unaddressed, results in unreliable, biased, and inflated TMB estimates. Here, we apply machine learning to the task of somatic vs germline classification in tumor-only solid tumor samples using TabNet, XGBoost, and LightGBM, three machine-learning models for tabular data. We constructed a training set for supervised classification using features derived exclusively from tumor-only variant calling and drawing somatic and germline truth labels from an independent pipeline using the patient-matched normal samples. All three trained models achieved state-of-the-art performance on two holdout test datasets: a TCGA dataset including sarcoma, breast adenocarcinoma, and endometrial carcinoma samples (AUC > 94%), and a metastatic melanoma dataset (AUC > 85%). Concordance between matched-normal and tumor-only TMB improves from R 2 = 0.006 to 0.71–0.76 with the addition of a machine-learning classifier, with LightGBM performing best. Notably, these machine-learning models generalize across cancer subtypes and capture kits with a call rate of 100%. We reproduce the recent finding that tumor-only TMB estimates for Black patients are extremely inflated relative to that of white patients due to the racial biases of germline databases. We show that our approach with XGBoost and LightGBM eliminates this significant racial bias in tumor-only variant calling. |
first_indexed | 2024-03-11T13:46:40Z |
format | Article |
id | doaj.art-c5b2cf3cf2c544e188a700aec68c1107 |
institution | Directory Open Access Journal |
issn | 2397-768X |
language | English |
last_indexed | 2024-03-11T13:46:40Z |
publishDate | 2023-01-01 |
publisher | Nature Portfolio |
record_format | Article |
series | npj Precision Oncology |
spelling | doaj.art-c5b2cf3cf2c544e188a700aec68c11072023-11-02T10:17:19ZengNature Portfolionpj Precision Oncology2397-768X2023-01-017111210.1038/s41698-022-00340-1Fast, accurate, and racially unbiased pan-cancer tumor-only variant calling with tabular machine learningR. Tyler McLaughlin0Maansi Asthana1Marc Di Meo2Michele Ceccarelli3Howard J. Jacob4David L. Masica5Genomics Research Center, AbbVieAgricultural and Biological Engineering at Purdue UniversityJohns Hopkins UniversityDepartment of Electrical Engineering and Information Technology, University of Naples “Federico II”Genomics Research Center, AbbVieGenomics Research Center, AbbVieAbstract Accurately identifying somatic mutations is essential for precision oncology and crucial for calculating tumor-mutational burden (TMB), an important predictor of response to immunotherapy. For tumor-only variant calling (i.e., when the cancer biopsy but not the patient’s normal tissue sample is sequenced), accurately distinguishing somatic mutations from germline variants is a challenging problem that, when unaddressed, results in unreliable, biased, and inflated TMB estimates. Here, we apply machine learning to the task of somatic vs germline classification in tumor-only solid tumor samples using TabNet, XGBoost, and LightGBM, three machine-learning models for tabular data. We constructed a training set for supervised classification using features derived exclusively from tumor-only variant calling and drawing somatic and germline truth labels from an independent pipeline using the patient-matched normal samples. All three trained models achieved state-of-the-art performance on two holdout test datasets: a TCGA dataset including sarcoma, breast adenocarcinoma, and endometrial carcinoma samples (AUC > 94%), and a metastatic melanoma dataset (AUC > 85%). Concordance between matched-normal and tumor-only TMB improves from R 2 = 0.006 to 0.71–0.76 with the addition of a machine-learning classifier, with LightGBM performing best. Notably, these machine-learning models generalize across cancer subtypes and capture kits with a call rate of 100%. We reproduce the recent finding that tumor-only TMB estimates for Black patients are extremely inflated relative to that of white patients due to the racial biases of germline databases. We show that our approach with XGBoost and LightGBM eliminates this significant racial bias in tumor-only variant calling.https://doi.org/10.1038/s41698-022-00340-1 |
spellingShingle | R. Tyler McLaughlin Maansi Asthana Marc Di Meo Michele Ceccarelli Howard J. Jacob David L. Masica Fast, accurate, and racially unbiased pan-cancer tumor-only variant calling with tabular machine learning npj Precision Oncology |
title | Fast, accurate, and racially unbiased pan-cancer tumor-only variant calling with tabular machine learning |
title_full | Fast, accurate, and racially unbiased pan-cancer tumor-only variant calling with tabular machine learning |
title_fullStr | Fast, accurate, and racially unbiased pan-cancer tumor-only variant calling with tabular machine learning |
title_full_unstemmed | Fast, accurate, and racially unbiased pan-cancer tumor-only variant calling with tabular machine learning |
title_short | Fast, accurate, and racially unbiased pan-cancer tumor-only variant calling with tabular machine learning |
title_sort | fast accurate and racially unbiased pan cancer tumor only variant calling with tabular machine learning |
url | https://doi.org/10.1038/s41698-022-00340-1 |
work_keys_str_mv | AT rtylermclaughlin fastaccurateandraciallyunbiasedpancancertumoronlyvariantcallingwithtabularmachinelearning AT maansiasthana fastaccurateandraciallyunbiasedpancancertumoronlyvariantcallingwithtabularmachinelearning AT marcdimeo fastaccurateandraciallyunbiasedpancancertumoronlyvariantcallingwithtabularmachinelearning AT michelececcarelli fastaccurateandraciallyunbiasedpancancertumoronlyvariantcallingwithtabularmachinelearning AT howardjjacob fastaccurateandraciallyunbiasedpancancertumoronlyvariantcallingwithtabularmachinelearning AT davidlmasica fastaccurateandraciallyunbiasedpancancertumoronlyvariantcallingwithtabularmachinelearning |